Why Flowstride?
In Flowstride (FS), a useful test tells the next person what the user tried to do, what happened, and how to reproduce a failure. Flowstride keeps those three concerns close together.
Describe intent in the flow
flow.click button "Log in" near "Sign up";
flow.type input "Email" "{{env.TEST_EMAIL}}";
flow.type input "Password" "{{env.TEST_PASSWORD}}";
flow.click button "Login";
flow.expect visible "Logout";Readable targets make reviews easier. Element qualifiers, CSS selectors, and spatial relationships let you become more specific when a page has repeated labels. You still need an application with stable, distinguishable controls.
Keep API setup and browser behavior together
Create data with an API, extract a returned identifier, then use the UI to exercise the customer journey. Both parts share extracted/generated values in the flow file, so you can verify the same entity through each interface. API and UI in one flow shows the complete journey.
When the assertion needs code
Suppose a catalog has 50 cars and three discount rules. Put the calculation in a JavaScript function. Pass the active browser page to a TypeScript helper that matches each displayed price by SKU. The scenario still tells a reviewer which catalog was fetched, which prices were compared and how many matched. Both helpers are ordinary named exports. Run the full example
Use one suite locally and in CI
Interactive runs provide step history, network records, Replay, and local debugging. Native CI runs the .flow suite without a custom runner file and produces machine-readable reports. A small generated GitHub workflow handles installation and evidence retention.
Give manual work a place in the process
Map a user journey, write observable expected results, execute it manually, and preserve the run with defects and evidence. A stable manual case can become an automated scenario later. Conversion is an authoring workflow; there is no automatic canvas-to-DSL converter.
Know the boundary
Flowstride uses Playwright for browser control, but does not expose every Playwright feature as a native DSL command. It supports Chromium, Firefox and WebKit, with native profiles, retrying assertions, network control, downloads, evidence, accessibility, component tests and browser-game APIs. Browser/device qualification still has explicit limits. See capabilities and limits for the distinction between native commands, JS/TS helpers and SDK plugins.
